Oak Mast Production and Animal Impacts on Acorn Survival in the Central Hardwoods

176

OAK MAST PRODUCTION AND ANIMAL IMPACTS ON ACORN SURVIVAL IN THE CENTRAL HARDWOODS

Kenneth F. Kellner, Jeffery K. Riegel, Nathanael I. Lichti, and Robert K. Swihart1

Abstract.—As part of the Hardwood Ecosystem Experiment we measured mast production in white (Quercus alba) and black (Q. velutina) oak, and quantified the impacts of seed predators on acorn survival over a 3-year period. Specifically, we measured the proportion of acorns of each species infested with weevils (Curculio spp.), and the probability of acorn removal by seed predators from a system of semipermeable exclosures. The 3 years of the study included 2 years of high mast production in both species (2006-07) and 1 year of mast crop failure (2008). Across all 3 years, 19 percent of acorns were infested. The rate of weevil infestation was slightly higher for black oak than for white oak in each year, but infestation peaked for both species during the year of mast failure (2008). Overall, 39 percent of acorns in the exclosures were removed by seed predators. The probability of acorn removal was lower when squirrels were excluded, providing support for additive effects of different seed predators. The probability of removal was highest in 2008 during the year of mast failure. In the future, these pre- harvest data will be compared to data obtained following timber harvests conducted in winter 2009.

INTRODUCTION 2001, Johnson et al. 2002). Among the most important Oak (Quercus L.) is a dominant overstory species functions of oak in the Central Hardwoods is the group in the Central Hardwoods Region. Oaks have production of hard mast, an important food source been labeled as both a keystone and “foundation” for at least 44 species of birds, small mammals, species in eastern deciduous forests (Ellison et al. and larger vertebrates such as white-tailed deer 2005, Fralish 2004), performing a wide variety (Odocoileus virginianus Zimmerman) and black bear of functions in forest ecosystems. For example, (Ursus americanus Pallus) (McShea et al. 2007). oak-dominated forests promote biodiversity in the The importance of oak as a source of hard mast has herbaceous understory because oak branch structure greatly increased in the past century due to the decline allows a large amount of sunlight to reach the forest of American chestnut (Castanea dentata [Marsh.] floor (Fralish 2004, Horn 1971). Oaks also are Borkh.) (Dalgleish and Swihart [in press], Diamond et known to provide habitat for many species of insects, al. 2000, McShea et al. 2007). fungi, and vertebrates and play an important role in hydrology and nutrient cycling (Brändle and Brandl Oaks generally are shade-intolerant species, requiring disturbance to regenerate effectively (Larsen and Johnson 1998). An active cycle of natural and/or 1 Graduate Student (KFK, NIL); Field Research Coordinator anthropogenic fire disturbance is thought to have (JKR); and Professor and Department Head (RKS), Purdue promoted oak dominance in the Central Hardwoods University, Department of Forestry and Natural Resources, before European settlement (Abrams 1992). Following 715 West State St., West Lafayette, IN 47907. KFK is corresponding author: to contact, call 765-494-9597 or settlement, cycles of land clearing for agriculture and email at [email protected]. subsequent abandonment maintained oak presence in 177

the overstory (Fralish 1997). However, the advent of 2000, McShea et al. 2007). However, some seed fire suppression and the creation of protected national predators (e.g., S. carolinensis) make many small and state parks in the 20th century have greatly caches of acorns, which may promote germination reduced the frequency of disturbance in eastern forests success (Barnett 1977, Smallwood et al. 2001, Steele (Abrams 1992, 2003). The result has been regeneration et al. 2006). failure of oaks across the Central Hardwoods Region, as well as shifts in species assemblages within the oak Regardless of animals’ influence on oak regeneration genus (Abrams 2003, Aldrich et al. 2005). This altered success, little is known about the impact of timber disturbance regime favors the establishment of shade- harvest strategies on oak mast production and tolerant climax species such as maple (Acer L.) and subsequent predation by insects and small mammals. American beech (Fagus grandifolia Ehrh.) (Fralish The Hardwood Ecosystem Experiment (HEE), a long- 2003). term replicated study in Indiana of forest ecosystem responses to timber harvest, provides an excellent Loss of oak as a canopy dominant would have experimental framework to address this knowledge important ecological and economic impacts (Johnson gap. Our objectives in this study were to compare mast et al. 2002, McShea et al. 2007). As a result, production by black (Q. velutina Lamb.) and white researchers have begun to develop forest management (Q. alba L.) oaks at several sites in southern Indiana (i.e., timber harvest) strategies to promote oak over a 3-year period, and subsequently to assess the regeneration (Dey and Parker 1996, Dey et al. 2008, impacts of acorn weevils and seed predators on the Loftis 1990). Examples of management include even- acorn crop. Following the final year of this preliminary aged methods (clearcuts and shelterwood harvests) study, the experimental sites were harvested under and uneven-aged methods (group-selection or single- several different management strategies with the goal tree selection harvests). Management efforts have of improving oak regeneration. Ultimately, our results met with mixed success; ecological variables such will serve as a baseline for identifying changes in mast as soil quality and moisture likely play a role in the production and seed predation by weevils and small outcome (Dey et al. 2009). Oak regeneration success mammals following the silvicultural treatments. is also heavily influenced by animals, including acorn weevils, small mammal seed predators, and deer Prior to the application of silvicultural treatments, (Marquis et al. 1976). we tested several hypotheses. We expected that acorn production would vary among the 3 years of the study, Weevils (Curculio L.) infest the acorns of all oak and that production would be different between the species in the Central Hardwoods. In a given year, as two oak species. Oaks are a masting species group, many as 50-90 percent of all acorns produced may be synchronizing with other trees in a region to produce infested (Gribko 1995, Lombardo and McCarthy 2008, very large or very small mast crops (Janzen 1971), but Marquis et al. 1976, Riccardi et al. 2004). Infested often lacking interspecific synchrony (Abrahamson acorns are both less likely to be dispersed (Steele et and Layne 2003, Lombardo and McCarthy 2008, Lusk al. 1996) and less likely to germinate successfully et al. 2007). We also expected that black and white (Andersson 1992, Lombardo and McCarthy 2009). oaks may be affected differently by acorn weevils; The interaction of oaks with small mammal seed Lombardo and McCarthy (2008) found that a higher predators is more complex. A large percentage of proportion of acorns in the red oak section (Lobatae) fallen acorns are eventually removed by seed predators were infested than those in the white oak section such as white-tailed deer, gray squirrels (Sciurus (Quercus), but few studies have compared infestation carolinensis Gmelin), eastern chipmunks (Tamius levels between oak species. striatus L.) and mice (Peromyscus Gloger) (McShea 178

To address the impacts of predators on the fate of Across all experimental units, 108 trees were sampled fallen seeds, we sought to isolate the individual in 2006, divided evenly among the three harvest contributions of several acorn predators on the total treatments. In 2007, an additional experimental amount of removals observed, in order to assess if unit was added in Brown County State Park; four those contributions were additive or compensatory. additional control trees were selected in this unit for We expected that gray squirrels would be most prolific a total of 112 trees sampled. Due to active timber at removing fallen acorns (Bellocq et al. 2005). In harvesting in uneven- and even-aged units in the fall of addition, we expected that less desirable acorns (i.e., 2008, 20 trees could not be sampled, yielding a sample acorns that were broken, germinated, or infested with size of 96 in the third year of the experiment. Overall, weevils) would be less likely to be taken by small black and white oaks had diameters at breast height mammals (Smallwood et al. 2001, Steele et al. 1996). (d.b.h.) of 20.0±4.4 cm (mean ± standard deviation) and 19.5±4.5 cm, respectively.

STUDY AREA AND DESIGN A summary of site characteristics and the selection MATERIALS and METHODS of research areas can be found in a previous chapter Mast Production of this report (Kalb and Mycroft, this publication). At each tree, two mast collection traps were Following the delineation of the research core areas, established. Traps consisted of a 52cm × 33cm × 32cm we selected mature black and white oak trees to plastic bin mounted atop a 2-m high polyvinyl chloride be included in the study. We chose this pair of oak pipe driven into the ground. Bins were covered with species because they are among the most dominant chicken wire to prevent pilferage of seeds by animals tree species in the region (Jenkins and Parker 1998), while still allowing mast to fall into the trap. Traps and to ensure that both oak sections were represented. were placed midway between the trunk and canopy Individual trees of reproductive age were chosen based edge underneath a limb. For trees adjacent to proposed on their location relative to future harvests. harvest areas, one mast trap was placed on the side of the tree facing the harvest and one on the opposite The number and location of selected trees depended side. For all other trees, one mast trap was oriented to on the assigned management treatment of the unit. In the north of the trunk and the other to the south. each of the three control units, six trees of each species were selected, with two near the center of the research In the first year of the study (2006), traps were set up core and four >100 m away near the core edge. In in October and checked five times at weekly intervals the three units assigned uneven-aged management, until the beginning of December. In 2007 and 2008, six trees of each species were selected, with all trees traps were set up in late August and checked eight adjacent to (but outside) the proposed 0.40-, 1.21-, times at 1- to 2-week intervals until December. Some and 2.02-ha patch cuts. In the even-aged units, the traps were not sampled in uneven-aged units in 2008 arrangement of trees differed between harvest types. due to ongoing timber harvests in the area. At each Each even-aged unit contained four trees of each trap check, all acorns in the collection buckets were species associated with shelterwood harvests; half counted, identified to species, and removed. were adjacent to the future shelterwoods and half were selected as overstory trees within the shelterwoods Weevil Infestation to be retained during the harvest. Each even-aged The majority of acorns removed from mast collection unit also contained two trees of each species located bins were examined for weevil infestation. In 2007 adjacent to clearcut areas. and 2008, a small number of additional acorns 179 were removed from the forest floor underneath hexagonally meshed chicken wire. The wire began 15 the mast trees to increase the sample size. Acorns cm above the ground to allow access to squirrels and were X-rayed to identify damage caused by weevils smaller wildlife. The third exclosure excluded squirrels (Dixon et al. 1997, Lombardo and McCarthy 2009). and larger wildlife but allowed access to mice and Acorns were marked according to source tree and chipmunks (C+M) and was a 0.75m × 0.75m × 0.2m collection date and X-rayed in groups of 50 using a wooden frame covered in 3.8-cm chicken wire. The Specimen Radiography X-ray System (Faxitron X- final control exclosure was designed to prevent access ray Corporation, Lincolnshire, IL). Weevil damage by all vertebrate wildlife (A) and was a wooden frame is indicated on the X-ray film by characteristic dark of the same size but covered instead with 0.6-cm mesh patterns within the acorn (Fig. 1; see Lombardo and hardware cloth. McCarthy 2009: Fig. 1). We classified an acorn as infested if any portion of the acorn interior exhibited The four exclosures were arranged randomly on a weevil damage. north-to-south transect bisecting the trunk of the mast study tree. Two exclosures were placed to the north Acorn Removal of the tree and two to the south. On each sampling A set of four semipermeable exclosures was occasion, acorns that had fallen into each exclosure established underneath each mast tree to assess the were located and numbered with a black marker. It was impacts of several groups of acorn predators. In impossible for acorns to fall naturally into the control the study forests, acorn predators include white- exclosures, so the A exclosures were provisioned with tailed deer, gray squirrels (and a few fox squirrels, a number of acorns equal to the average number inside S. niger L.), eastern chipmunks, and white-footed the other three. On subsequent visits, the presence mice (P. leucopus). Each exclosure was a 0.75m × or absence of the marked acorns was determined. 0.75m square. The first exclosure allowed access to If an acorn was present, information about the all wildlife species (deer, squirrels, chipmunks, and acorn including the presence of weevil exit holes, mice or D+S+C+M) and simply consisted of four germination status, and integrity of the acorn (broken wooden stakes designating a 0.75m × 0.75m area on or intact) was recorded. the ground. The second excluded deer only (allowing access to S+C+M) and consisted of four wooden Analysis stakes covered on the sides and top by 3.8-cm wide Mast Production The total number (count) of acorns collected at a given tree during a given sampling occasion was the response variable in this analysis. Overdispersion of these count data, with a variance ~5 times greater than the mean value, prevented analysis with simple Poisson regression. Instead, we fit a Poisson- lognormal model (Kéry 2010) which introduced an additional free parameter to allow the variance of the distribution to differ from the mean. Briefly, we

modeled Ci,j,k , the observed acorn count at tree i during

sampling period j of year k as Ci,j,k ~ Poisson(λi,j,k ). The linear predictor for the Poisson intensity parameter Figure 1.—X-ray of two black oak acorns; the acorn on the left is infested with weevil larvae and the acorn on the right λ was λi,j,k = exp(βqXq,ijk + εi,j,k ) where β is a vector of is sound. slope parameters, X is a matrix of q covariate values, 180

and ε is distributed N(0, σ2), to allow the variance to parameters were assigned a normal prior distribution vary independently from λ. We included covariates for with mean 0 and variance 1,000, and variance tree species, d.b.h., the length and date of sampling parameters were assigned a lognormal distribution periods (in Julian day format), and proposed treatment with mean 0 and variance 1 following Kéry (2010). in the linear predictor. In addition to the random error All covariates were standardized by subtracting ε, we considered tree and year to be random effects. the mean and dividing by the standard deviation. Bayesian analyses were conducted using the software Weevil Infestation package WinBUGS (version 1.4.3; Cambridge, UK) We estimated the effects of acorn-specific covariates (Spiegelhalter et al. 1996), which uses the Markov Chain Monte Carlo (MCMC) method to generate on the infestation status zi of individual collected posterior distributions for the model parameters. acorns. Infestation zi was modeled as zi | pi ~ WinBUGS was called from within R 2.10.0 (R Bernoulli(pi ), where logit(pi ) is a linear function of covariates. The linear predictor included year, Julian Foundation for Statistical Computing, Vienna, Austria) day of sampling, Julian day of sampling centered using the R2WinBUGS library (Sturtz et al. 2005). squared, proposed management technique for the source site of acorn i, species of acorn i as fixed MCMC simulations were conducted with a minimum effects, and source tree as a random effect. of 3 chains, 30,000 iterations, and a burn-in of 10,000 iterations. Convergence was assessed by calculating Acorn Removal the Gelman-Rubin diagnostic Rhat (Brooks and We related site- and acorn-specific covariates to Gelman 1998) for each estimate parameter. If we did not observe convergence in the distributions of all the probability Φi,j,k that acorn i would be removed between sampling occasions j-1 and j in year k. The posterior parameters (Rhat > 1.1), MCMC chain length and burn-in were increased until convergence was presence zi,j,k of a given acorn i in an exclosure at sampling occasion j in year k took on a value of 0 or 1 achieved. Posterior predictive checks were performed and was modeled conditional on the acorn’s state at on each regression model to assess goodness of fit; for the mast production model Pearson’s chi-square the previous sampling occasion zi,j-1,k and probability residuals were compared between the actual data and of removal Φi,j,k ; that is, zi,j,k | zi,j-1,k , Φi,j,k ~ data simulated based on the model, and for the other Bernoulli(Φi,j,k ). Logit(Φi,j,k ) was set equal to a linear predictor containing acorn- and site-specific two analyses the sums of absolute residuals were covariates. Explanatory variables considered in compared. Bayesian p-values were calculated for each this analysis were the fixed effects of Julian day posterior predictive check. An individual covariate of sampling for occasion j, Julian day of sampling was considered to have an important effect if the 95- centered squared, exclosure type for acorn i, percent credible interval of its slope parameter did not proposed management treatment for acorn i, and the contain zero. germination, weevil infestation, and shell status (i.e., broken or intact) of acorn i at time j-1. Length of time between sampling periods j-1 and j was considered RESULTS as a nuisance variable, and source tree and year were Mast Production included as random effects. We collected 1,140 acorns over the 3 years of the experiment (Table 1). The Poisson-lognormal model Model Fitting had acceptable fit; a posterior predictive check yielded All regression models were fit in a Bayesian a Bayesian p-value of 0.41 (0.5 is ideal). framework with non-informative priors. Slope 181

Table 1.—Number of trees sampled and acorn sample sizes for the mast production, weevil infestation, and mast removal experiments in 2006-08.

Parameter 2006 2007 2008 Total

Trees sampled 108 112 96 Total acorns collected in traps 355 670 115 1,140 Black oak (Q. velutina) 235 414 9 658 White oak (Q. alba) 120 256 106 482 Total acorns x-rayed for weevils 366 595 114 1,075 Black oak 265 423 8 696 White oak 101 172 106 379 Total acorns monitored for removal 1,041 1,827 240 3,108 Black oak 468 1,013 30 1,511 White oak 573 814 210 1,597

In general, white oaks produced fewer acorns than black oaks in our study area (Fig. 2), except in 2008, when there was a mast failure in black oak. The 95- percent credible interval for the species coefficient in the Poisson regression included zero (Table 2). However, 74 percent of the posterior distribution of the species coefficient was below zero, providing marginal support for lower production by white oak. Production by individual trees was positively correlated with d.b.h.; larger trees produced more acorns. Counts were negatively correlated with both Julian day of sampling and Julian day squared, indicating that production was highest early in the sampling period (September- October), declining later in the season. Proposed Figure 2.—Total acorns collected by species and year. Since harvest treatment had no effect on acorn production sampling began in October 2006, only acorns collected in the months of October through December are included to among the trees we sampled. allow direct comparisons between years.

Table 2.—Estimated parameter values for the Poisson-lognormal model of oak mast production. The value f represents the proportion of the posterior distribution for the parameter which has the same sign (positive or negative) as the mean. Parameters with 95-percent credible intervals which do not include zero are marked with an asterisk (*).

Parameter Covariate type Mean SE f

Intercept -3.21 0.98 1.00 Tree species (WO=1) indicator -0.17 0.26 0.74 d.b.h. continuous 0.33* 0.13 0.99 Sample day (Julian) continuous -1.06* 0.12 1.00 Julian day squared continuous -1.75* 0.13 1.00 Even-aged treatment effect indicator -0.46 0.31 0.93 Uneven-aged treatment effect indicator 0.12 0.32 0.64 182

Weevil Infestation Infestation probability increased from 2006 to 2008 A total of 1,075 acorns were X-rayed to identify (Table 3). The high probabilities of infestation in weevil infestation (Table 1). The number of X-rayed both species in 2008 (Fig. 3) coincided with low mast acorns does not correspond exactly to the number production in white oak and a mast failure in black removed from buckets because small numbers of oak in 2008 (Fig. 2). In addition to the year effects, acorns were lost following collection and/or added there was a negative correlation between Julian day from the ground beneath the bucket. A posterior of sampling and probability of infestation (Table 3); predictive check of the logistic regression model fit acorns collected later in the fall were less likely to to the data yielded an acceptable Bayesian p-value have weevil damage. There were no differences in of 0.39. Black oak acorns had a higher probability of weevil infestation between areas selected to receive infestation in all three sample years compared to white different silvicultural treatments. oak (Fig. 3, Table 3). Acorn Removal Over the 3-year study period, 3,108 acorns were marked and monitored for removal (Table 1). Of these, 1,200 (39 percent) were removed from the exclosures at some point: 595 of 1,511 black oak acorns (39 percent) and 605 of 1,597 white oak acorns (38 percent). We used logistic regression to model the probability that an individual acorn was removed during a given time period [ j-1, j]. Our model fit the data adequately, with a posterior predictive check of absolute residuals yielding a Bayesian p-value of 0.38.

There were no differences in the probability of removal between acorn species. However, acorn damage, weevil infestation, and germination were all negatively correlated with the probability of acorn Figure 3.—Posterior distributions for the mean probabilities removal (95-percent credible intervals excluding zero; of acorn infestation, by acorn species and year. Error bars Table 4). There were no differences in probability of represent 95-percent credible intervals. removal between proposed treatment areas.

Table 3.—Estimated parameter values for the model of probability of acorn infestation. The value f represents the proportion of the posterior distribution for the parameter which has the same sign (positive or negative) as the mean. Parameters with 95-percent credible intervals which do not include zero are marked with an asterisk (*).

Parameter Covariate type Mean SE f

Intercept -2.02 0.26 1.00 2007 year effect indicator 0.72* 0.22 1.00 2008 year effect indicator 1.96* 0.34 1.00 Acorn species (WO=1) indicator -0.52* 0.25 0.98 Sample day (Julian) continuous -0.18* 0.094 0.98 Julian day squared continuous 0.015 0.084 0.58 Even-aged treatment effect indicator -0.084 0.27 0.62 Uneven-aged treatment effect indicator -0.071 0.26 0.62 183

Table 4.—Estimated parameter values for the model of probability of acorn removal by predators. The value f represents the proportion of the posterior distribution for the parameter which has the same sign (positive or negative) as the mean. Parameters with 95-percent credible intervals which do not include zero are marked with an asterisk (*).

Parameter Covariate type Mean SE f

Intercept -4.34 0.60 0.99

Date and harvest treatment parameters Sample day (Julian) continuous 0.049 0.055 0.82 Julian day squared continuous 0.24* 0.065 1.00 Time between samples continuous -0.055 0.038 0.92 Even-aged treatment effect indicator 0.12 0.26 0.67 Uneven-aged treatment effect indicator -0.011 0.27 0.52

Acorn characteristics Acorn species (WO=1) indicator -0.082 0.22 0.66 Acorn integrity (interact=1) indicator -0.48* 0.15 1.00 Germination status (germ=1) indicator -0.49* 0.15 1.00 Weevil status (infested=1) indicator -0.78* 0.21 1.00

Exclosure-type parameters (baseline = completely inaccessible exclosure) Squirrel and larger exclosure indicator 3.39* 0.24 1.00 Deer and larger exclosure indicator 4.09* 0.24 1.00 Open exclosure indicator 3.91* 0.24 1.00

Acorns in all three exclosures accessible to at least some vertebrates (open, deer, and squirrel exclosures) had a higher probability of removal than acorns in the control, inaccessible exclosure (Table 4, Fig. 4). Differences between the three accessible exclosure types were more difficult to ascertain. The posterior distributions for the exclosure regression coefficients overlapped (Fig. 5), with the coefficient corresponding to squirrel exclosures having a smaller mean value, corresponding to a lower probability of removal.

In 2007 and 2008, the 95-percent credible interval for removal probability from squirrel exclosures did Figure 4.—Mean probabilities of removal from each not overlap with the open or deer exclosures credible exclosure type (for an individual sampling period) during the 3 years of the study. Exclosure O was open to all animals, intervals, providing further evidence that excluding D excluded deer, S excluded everything squirrel-sized and access to squirrels reduced the probability of acorn larger, and A (the control) excluded all vertebrates. Error bars removal (Fig. 4). In the same years, probability of represent 95-percent credible intervals. removal was higher from the exclosures that excluded only deer. In general, probabilities of removal were higher in 2008, corresponding to reduced mast production (Figs. 2 and 4). 184

acorns between these species. In addition, our study did not incorporate population-level information on oaks in the study area; consequently, we cannot make conclusions about the overall importance of the two oak species at our sites.

Though white and black oaks are common at our study sites, it is important to note that the total amount of mast available also depends on other hardwood species (especially chestnut oak, but also Northern red oak, Q. rubra, and the hickories, Carya). We did not monitor mast production by these species but because there is often a lack of interspecific synchrony in Figure 5.—Comparison of posterior distributions for the three exclosure-type parameters. The three parameter values masting (Abrahamson and Layne 2003, Lombardo and are relative to removal from a control, i.e., a completely McCarthy 2008, Lusk et al. 2007), they may buffer the inaccessible exclosure. total mast crop available for wildlife in the study area.

Weevil Infestation DISCUSSION We found higher rates of weevil infestation in black Mast Production oak than in white oak in each year of the study (Fig. This 3-year study included 2 years of abundant mast 3), and overall there was a significant species effect for (2006-07) and 1 year of mast failure in black oak white oak (Table 3)). Our results are consistent with and partial mast failure in white oak (2008; Fig. 2). the findings of Lombardo and McCarthy (2008), who More black oak acorns were collected per sampling found higher rates of infestation in black oak than in occasion (traps × sample periods) in the non-failure chestnut oak (which belongs to the white oak section) years, indicating that individual black oaks in our in southeastern Ohio. Black oak may be a more study region may be more valuable than white oaks for consistent mast producer than chestnut oak (Lombardo wildlife food production. This result is consistent with and McCarthy 2008) and white oak (Sork and Bramble other studies that found black oak to generally be a 1993, Sork et al. 1993), which may allow the black larger producer of acorns than chestnut oak (Q. prinus) oak population to support a larger number of weevils. and white oak (Lombardo and McCarthy 2008, Sork A limitation of this study is that weevils were not and Bramble 1993, Sork et al. 1993). identified to the species level, and different members of the genus Curculio appear to specialize on certain Overall, however, our model did not identify a species oak species (Gibson 1972, 1982). It is possible that the effect that was statistically different from zero (Table particular assemblage of individual Curculio species at 2). In the short duration of our study, production our study sites is responsible for higher infestation in by black oak appeared to be more variable than black oak. Nearly all acorns were collected from raised white oak (Fig. 2), though the literature suggests buckets, so we do not believe many of the acorns were it should be a consistent producer (Lombardo and infested with weevils in the genus Conotrachelus McCarthy 2008, Sork and Bramble 1993, Sork et al. (Schonherr), which generally lays eggs in previously 1993). A longer record of production is necessary infested or broken acorns that have already fallen to to accurately compare variability in production of the ground (Gibson 1964). 185

While there were year effects for both 2007 and 2008 squirrels, the primary dispersal agent in our system, (Table 3), infestation rates were highest for both are sensitive to acorn condition and are less likely to species in 2008 (Fig. 3), corresponding to a mast cache infested acorns (Steele et al. 1996). In addition, failure in black oak and a reduced mast year in white because white oak begins to germinate in the fall, oak (Fig. 2) across the study sites. The previous 2 squirrels are less likely to cache them for later use years were good mast years for both white and black than acorns from the red oak section (Smallwood et al. oak (Fig. 2), likely resulting in a large population of 2001). The most likely explanation for this behavior Curculio larvae undergoing diapause in the soil. When is that broken, infested, and germinated acorns are this population emerged from the soil in fall 2008, more perishable and therefore less valuable to store only a few acorns were available, resulting in very for the winter months. Infested or broken acorns high rates of infestation. Lombardo and McCarthy may still be removed and eaten immediately by the (2008) identified a similar pattern in chestnut oak in predator (Steele et al. 1996), but in years when mast southeastern Ohio. They found 86 percent of chestnut is plentiful (2006-2007 in this study), predators may oak acorns were predated by weevils in a non-mast ignore less desirable acorns that are not intact. In the year, but only 26 percent were infested in a mast year. year of mast failure (2008), removal probability during Unlike our study, however, they found black oak to a given sample period was very high (Fig. 4), and have relatively consistent weevil predation regardless nearly all fallen acorns were removed by the end of the of the quantity of mast produced. experiment regardless of their condition. The design of this experiment did not allow us to ascertain the fate The mean probabilities of black oak acorn infestation of acorns post-removal, so we were unable to identify that we observed across the 3 years of the study (0.12- differences in the ultimate fate of acorns of varying 0.48 for black oak, and 0.08-0.38 for white oak) were species and conditions. generally lower than in similar studies. For example, Lombardo and McCarthy (2008) reported that on The system of semipermeable exclosures used in this average 79 percent of black oaks were infested over a study allowed us to identify the impacts of several 4-year period. Riccardi et al. (2004) found that 55 and groups of acorn predators on overall removal rates. 34 percent of black oaks were infested, respectively, When deer and turkeys (Meleagris gallopavo) were in the 2 years of their study. The lower numbers we excluded, removal probabilities actually increased report may be due to differences in the assemblage of in each year (compare exclosures O and D in Figure acorn weevil species between sites. Differences in host 4), though differences between the two exclosures tree and weevil predator densities may also have an were not different from zero (Table 4). Posterior impact on weevil populations. For example, Anderson distributions for the slope parameters of the deer and and Folk (1993) found that small mammal predators, open exclosure indicator variables overlapped almost especially white-footed mice and short-tailed shrews completely (Fig. 5). Therefore, large animals like deer (Blarina brevicauda Say) reduced overwinter survival do not appear to be an important predator of acorns of acorn weevils in Indiana. in this system as squirrels and smaller predators completely compensate for their effect when they are Acorn Removal by Predators excluded. However, any predation by deer is a dead Overall, 39 percent of acorns were eventually removed end for acorns, whereas removal by squirrels and other from the exclosures across the 3 years of the study. small mammals could result in either consumption or Seed predators did not seem to preferentially remove caching and later germination. Bellocq et al. (2005) either acorn species, but avoided acorns that were also observed that removal rates were higher when damaged, weevil-infested, or germinated (Table deer were excluded than when they had access to the 4). Previous studies have demonstrated that gray acorns, further evidence that deer are not an important 186

seed predator in our forest system. Increased rates production in trees at future even-aged treatment of removal when deer are excluded appear to be stands. Unfortunately, we are unable to scale our counter-intuitive; fewer animals in total have access results up from tree-level to site-level comparisons to the acorns. One possible explanation is that the because we did not account for population-level deer exclosure itself provided shelter from airborne differences in oak between sites in this study. Still, predators, increasing the probability that squirrels and these results provide a basis for attributing future mice would forage inside the exclosure. observed differences in tree-level mast production and predation to silvicultural treatments, rather than to pre- We did not observe a statistically significant difference existing site characteristics. in removal probability when squirrels were excluded (Table 4), but the trend (Fig. 4) indicates that removal We anticipate that timber harvest with the goal of probability was reduced; the posterior distributions regenerating oak will increase acorn production. for the slope parameters corresponding to the deer Harvesting at the clearcut and patch cut sites will and squirrel exclosures did not overlap greatly increase light penetration to oaks on the edge of (Fig. 5). Differences in small mammal community the cut, which will increase branch density and composition and demographics (e.g., very high mouse therefore acorn production (Johnson 1994, Verme and chipmunk populations) could be the reason we did 1953). However, these acorns play a minimal role in not observe an additive effect of squirrel predation as regeneration inside the clearcut (4.05 ha) and patch strong as the one observed by Bellocq et al. (2005). cut (0.40, 1.21, and 2.02 ha) harvest sites. In general, the primary dispersal agents in this system (small Treatment Effects and Predictions mammals and the blue jay, Cyanocitta cristata L.) do One of the primary objectives of this study was to not move acorns deep into recently harvested areas identify any differences in tree-level mast production (Nixon et al. 1980). More importantly, any seedlings and predation between HEE sites prior to applying the that do develop from dispersed acorns following silvicultural treatments. In general, we did not observe harvest will not be competitive without silvicultural many differences between trees in areas with different intervention (Sander 1972). The primary source of future management treatments. The parameters regeneration in the harvested areas will be seedlings of corresponding to proposed uneven- and even-aged sufficient size (>1.4 m tall) established prior to harvest treatment effects on probability of weevil infestation and stump sprouts (Sander 1972, Sander et al. 1984). and probability of acorn removal were not statistically different from zero (Tables 3 and 4). Mast production Dominant and codominant oaks are retained in the was similar between trees in future control, uneven-, overstory during the first phase of the shelterwood and even-aged management sites (Table 2). However, method for regeneration (Loftis 1990). Removal of the tree-level mast production was slightly lower at midstory and some competing trees should increase future even-aged sites (95 percent of the posterior light availability for oaks in the overstory, increasing distribution for the even-aged treatment parameter was acorn production (Johnson 1994). Evidence in the less than zero). literature is inconclusive. Thinning had a positive effect on red oak acorn production in New England Mast production in oaks can be highly variable from (Healy 1997) but had minimal effects on chestnut year to year (Sork et al. 1993), so it is possible that and black oak production in Ohio (Lombardo and observed differences in production between trees are McCarthy 2008). Bellocq et al. (2005) found that due to the short duration of the study. All sites at the production of red oak in Ontario appeared to increase HEE are relatively close together so it is unlikely when visual estimates were used, but no differences that weather or site quality is responsible for lower between treatments were observed in the number of 187 acorns collected in traps. Responses in production LITERATURE CITED to shelterwood harvests may be species-specific Abrahamson, W.G.; Layne, J.N. 2003. Long-term (Lombardo and McCarthy 2008); in addition, any patterns of acorn production for five oak species increases in production due to increased light may in xeric Florida uplands. Ecology. 84(9): have begun after the time period following harvest 2476-2492. covered by these studies. Abrams, M.D. 1992. Fire and the development of Currently, little is known about the effects of harvest oak forests. BioScience. 42(5): 346-353. treatments on vertebrate and invertebrate acorn Abrams, M.D. 2003. Where has all the white oak predators. Thinning and shelterwood harvests do not gone? BioScience. 53(10): 927-939. appear to change weevil infestation rates (Bellocq et al. 2005, Lombardo and McCarthy 2008). Weevil Aldrich, P.R.; Parker, G.R.; Romero-Severson, infestation of acorns produced by trees on the edge of J.; Michler, C.H. 2005. Confirmation of oak clearcuts and patch cuts is also unlikely to be affected. recruitment failure in Indiana old-growth forest: Predation by small mammals following harvest likely 75 years of data. Forest Science. 51(5): 406-416. depends on population levels relative to the amount Anderson, D.C.; Folk, M.L. 1993. Blarina brevicauda of mast available. Bellocq et al. (2005) found no and Peromyscus leucopus reduce overwinter differences in rates of acorn removal by predators survivorship of acorn weevils in an Indiana between shelterwoods and control stands, but there hardwood forest. Journal of Mammalogy. 74(3): were also no treatment effects on the abundance of 656-664. small mammals. A limitation of this and previous studies is that the ultimate fate of removed acorns Andersson, C. 1992. The effect of weevil and fungal was not determined. Future work should focus on attacks on the germination of Quercus robur identifying differences in seed fate between treatments, acorns. Forest Ecology and Management. 50(3-4): since removal could have either a positive (if the acorn 247-251. is cached and forgotten) or negative (if the acorn is Barnett, R.J. 1977. The effect of burial by squirrels eaten) effect on seedling establishment. on germination and survival of oak and hickory nuts. The American Midland Naturalist. 98(2): 319-330. ACKNOWLEDGMENTS Bellocq, M.I.; Jones, C.; Dey, D.C.; Turgeon, This paper is a contribution of the Hardwood J.J. 2005. Does the shelterwood method to Ecosystem Experiment, a partnership of the Indiana regenerate oak forests affect acorn production Department of Natural Resources (IDNR), Purdue and predation? Forest Ecology and Management. University, Ball State University, Indiana State 205(1-3): 311-323. University, Drake University, Indiana University of Pennsylvania, and The Nature Conservancy. Funding Brändle, M.; Brandl, R. 2001. Species richness for the project was provided by the IDNR-Division of insects and mites on trees: expanding of Forestry and Purdue University, Department of Southwood. Journal of Animal Ecology. 70(3): Forestry and Natural Resources. Numerous technicians 491-504. helped with data collection and entry. Cortney Mycroft Brooks, S.P.; Gelman, A. 1998. General methods for and Rebecca Kalb provided technical and logistical monitoring convergence of iterative simulations. assistance. Comments by M.R. Saunders and two Journal of Computational and Graphical Statistics. anonymous reviewers strengthened the manuscript. 7(4): 424-455. 188

Dalgleish, H.J.; Swihart, R.K. [In press]. American Fralish, J.S. 1997. Community succession, diversity, chestnut past and future: implications for and disturbance in the Central Hardwood resource pulses and consumer populations of Forest. In: M.W. Schwartz, ed. Conservation in eastern U.S. forests. Restoration Ecology. DOI: highly fragmented landscapes. New York: Chapman 10.1111/j.1526-00X.2011.00795.x. and Hall: 234-266.

Dey, D.C.; Jacobs, D.; McNabb, K.; Miller, Fralish, J.S. 2003. The Central Hardwood Forest: G.; Baldwin, V.; Foster, G. 2008. Artificial its boundaries and physiographic provinces. In: regeneration of major oak (Quercus) species Van Sambeek, J.W.; Dawson, J.O.; Ponder, F., Jr.; in the eastern United States – a review of the Loewenstein, E.F.; Fralish, J.S., eds. Proceedings, literature. Forest Science. 54(1): 77-106. 13th Central Hardwood Forest conference; 2002 April 1-3; Urbana, IL. Gen. Tech. Rep. NC-234. St. Dey, D.C.; Parker, W.C. 1996. Regeneration of Paul, MN: U.S. Department of Agriculture, Forest red oak (Quercus rubra L.) using shelterwood Service, North Central Research Station: 1-20. systems: ecophysiology, silviculture and management recommendations. Forest Research Fralish, J.S. 2004. The keystone role of oak and Information Paper Issue. 126. Sault Ste. Marie, ON: hickory in the Central Hardwood Forest. In: Ontario Forest Research Institute. 59 p. Spetich, M.A., ed. Upland oak ecology symposium: history, current conditions, and sustainability. Dey, D.C.; Spetich, M.A.; Weigel, D.R.; Johnson, Gen. Tech. Rep. SRS-73. Asheville, NC: U.S. P.S.; Graney, D.L.; Kabrick, J.M. 2009. A Department of Agriculture, Forest Service, suggested approach for design of oak (Quercus Southern Research Station: 78-87. L.) regeneration research considering regional differences. New Forests. 37(2): 123-135. Gibson, L.P. 1964. Biology and life history of acorn- infesting weevils of the genus Conotrachelus Diamond, S.J.; Giles, R.H., Jr.; Kirkpatrick, R.L.; (Coleoptera: Curculionidae). Annals of the Griffin, G.J. 2000. Hard mast production before Entomological Society of America. 57(5): 521-526. and after the chestnut blight. Southern Journal of Applied Forestry. 24(4): 196-201. Gibson, L.P. 1972. Insects that damage white oak acorns. Res. Pap. NE-220. Upper Darby, PA: Dixon, M.D.; Johnson, W.C.; Adkisson, C.S. 1997. U.S. Department of Agriculture, Forest Service, Effects of weevil larvae on acorn use by blue Northeastern Forest Experiment Station. 7 p. jays. Oecologia. 111(2): 201-208. Gibson, L.P. 1982. Insects that damage northern Ellison, A.M.; Bank, M.S.; Clinton, B.D.; Coblurn, red oak acorns. Res. Pap. NE-492. Broomall, PA: E.A.; Elliott, K.; Ford, C.R.; Foster, D.R.; Kloeppel, U.S. Department of Agriculture, Forest Service, B.D.; Knoepp, J.D.; Lovett, G.M.; Mohan, Northeastern Forest Experiment Station. 6 p. J.; Orwig, D.A.; Rodenhouse, N.L.; Sobczak, W.V.; Stinson, K.A.; Stone, J.K.; Swan, C.M.; Gribko, L.S. 1995. The effect of acorn insects on Thompson, J.; Von Holle, B.; Webster, J.R. 2005. the establishment and vigor of northern red Loss of foundation species: consequences for the oak seedlings in north-central West Virginia. structure and dynamics of forested ecosystems. In: Gottschalk, K.W.; Fosbroke, S.L.C., eds. Frontiers in Ecology and the Environment. 3(9): Proceedings, 10th Central Hardwood Forest 479-486. conference; 1995 March 5-8; Morgantown, WV. Gen. Tech. Rep. NE-197. Radnor, PA: U.S. Department of Agriculture, Forest Service, Northeastern Forest Experiment Station: 430-441. 189

Healy, W.M. 1997. Thinning New England oak Lusk, J.L.; Swihart, R.K.; Goheen, J.R. 2007. stands to enhance acorn production. Northern Correlates of interspecific synchrony and Journal of Applied Forestry. 14(3): 152-156. interannual variation in seed production by deciduous trees. Forest Ecology and Management. Horn, H.S. 1971. The adaptive geometry of trees. 242(2-3): 656-670. Princeton, NJ: Princeton University Press. 144 p. Marquis, D.A.; Eckert, P.L.; Roach, B.A. 1976. Acorn Janzen, D.H. 1971. Seed predation by animals. weevils, rodents, and deer all contribute to oak- Annual Review of Ecology and Systematics. regeneration difficulties in Pennsylvania. Res. 2: 465-492. Pap. NE-356. Upper Darby, PA: U.S. Department Jenkins, M.A.; Parker, G.R. 1998. Composition and of Agriculture, Forest Service, Northeastern Forest diversity of woody vegetation in silvicultural Experiment Station. 5 p. openings of southern Indiana forests. Forest McShea, W.J. 2000. The influence of acorn crops on Ecology and Management. 109(1-3): 57-74. annual variation in rodent and bird populations. Johnson, P.S. 1994. How to manage oak forests for Ecology. 81(1): 228-238. acorn production. Tech. Brief NC-1. St. Paul, MN: McShea, W.J.; Healy, W.M.; Devers, P.; Fearer, U.S. Department of Agriculture, Forest Service, T.; Koch, F.H.; Stauffer, D.; Waldon, J. 2007. North Central Forest Experiment Station. 2 p. Forestry matters: Decline of oaks will impact Johnson, P.S.; Shifley, S.R.; Rogers, R. 2002. The wildlife in hardwood forests. Journal of Wildlife ecology and silviculture of oaks. Wallingford, Management. 71(5): 1717-1728. Oxon, UK: CABI Publishing. 503 p. Nixon, C.M.; McClain, M.W.; Donohoe, R.W. 1980. Kéry, M. 2010. Introduction to WinBUGS for Effects of clear-cutting on gray squirrels. Journal ecologists. Burlington, MA: Elsevier Academic of Wildlife Management. 44(2): 403-412. Press. 277 p. Riccardi, C.L.; McCarthy, B.C.; Long, R.P. 2004. Larsen, D.R.; Johnson P.S. 1998. Linking the ecology Oak seed production, weevil (Coleoptera: of natural oak regeneration to silviculture. Forest Curculionidae) populations, and predation Ecology and Management. 106(1): 1-7. rates in mixed-oak forests of southeast Ohio. In: Yaussy, D.A.; Hix, D.M.; Long, R.P.; Goebel, Loftis, D.L. 1990. A shelterwood method P.C., eds. Proceedings, 14th Central Hardwood for regenerating red oak in the southern Forest conference; 2004 March 16-19; Wooster, Appalachians. Forest Science. 36(4): 917-929. OH. Gen. Tech. Rep. NE-316. Newtown Square, Lombardo, J.A.; McCarthy, B.C. 2008. Silvicultural PA: U.S. Department of Agriculture, Forest Service, treatment effects on oak seed production and Northeastern Research Station: 10-20. predation by acorn weevils in southeastern Ohio. Sander, I.L. 1972. Size of oak advance reproduction: Forest Ecology and Management. 255(7): key growth following harvest cutting. Res. 2566-2576. Pap. NC-79. St. Paul, MN: U.S. Department of Lombardo, J.A.; McCarthy, B.C. 2009. Seed Agriculture, Forest Service, North Central Forest germination and seedling vigor of weevil- Experiment Station. 6 p. damaged acorns of red oak. Canadian Journal of Forest Research. 39(8): 1600-1605. 190

Sander, I.L.; Johnson, P.S.; Rogers, R. 1984. Steele, M.A.; Hadj-Chikh, L.Z.; Hazeltine, J. 1996. Evaluating oak advance reproduction in the Caching and feeding decisions by Sciurus Missouri Ozarks. Res. Pap. NC-251. St. Paul, MN: carolinensis: responses to weevil-infested acorns. U.S. Department of Agriculture, Forest Service, Journal of Mammalogy. 77(2): 305-314. North Central Forest Experiment Station. 16 p. Steele, M.A.; Manierre, S.; Genna, T.; Contreras, Smallwood, P.D.; Steele, M.A.; Faeth, S.H. 2001. T.A.; Smallwood, P.D.; Pereira, M.E. 2006. The The ultimate basis of the caching preferences innate basis of food-hoarding decisions in grey of rodents, and the oak-dispersal syndrome: squirrels: evidence for behavioural adaptations tannins, insects, and seed germination. American to the oaks. Animal Behaviour. 71(1): 155-160. Zoologist. 41(4): 840-851. Sturtz, S.; Ligges, U.; Gelman, A. 2005. Sork, V.L.; Bramble, J.E. 1993. Prediction of acorn R2WinBUGS: a package for running WinBUGS crops in three species of North American oaks: from R. Journal of Statistical Software. 12(3): Quercus alba, Q. rubra and Q. velutina. Annals of 1-16. Forest Science. 50(supp.): 128s-136s. Verme, L.J. 1953. Production and utilization of Sork, V.L.; Bramble, J.; Sexton, O. 1993. Ecology of acorns in Clinton County, Michigan. East mast-fruiting in three species of North American Lansing, MI: Michigan State University. 77 p. deciduous oaks. Ecology. 74(2): 528-541. M.S. thesis.

Spiegelhalter, D.J.; Thomas, A.; Best, N.; Gilks, W. 1996. BUGS 0.5: Bayesian inference using Gibbs sampling manual (version ii). Cambridge, UK: MRC Biostatistics Unit. 59 p.

The content of this paper reflects the views of the author(s), who are responsible for the facts and accuracy of the information presented herein.